Microarrays of Tagged Combinatorial Triazine Libraries in the

A novel and highly diverse tagged triazine library incorporating a triethylene glycol-based linker was synthesized using an orthogonal combinatorial a...
0 downloads 5 Views 204KB Size
862

J. Comb. Chem. 2004, 6, 862-868

Articles Microarrays of Tagged Combinatorial Triazine Libraries in the Discovery of Small-Molecule Ligands of Human IgG Mahesh Uttamchandani,†,⊥ Daniel P. Walsh,‡,⊥ Sonya M. Khersonsky,‡ Xuan Huang,§ Shao Q. Yao,*,†,§ and Young-Tae Chang*,‡ Department of Biological Sciences and Department of Chemistry, National UniVersity of Singapore, 3 Science DriVe 3, Singapore 117543, Department of Chemistry, New York UniVersity, New York, New York 10003 ReceiVed June 10, 2004 A novel and highly diverse tagged triazine library incorporating a triethylene glycol-based linker was synthesized using an orthogonal combinatorial approach on the solid phase and covalently immobilized on a glass substrate as a small molecule microarray (SMM). The SMM was screened with a fluorophoreconjugated human IgG, and 4 novel binders from a library of 2688 compounds were identified from the fully spatially addressable array without the need for compound decoding. Using surface plasmon resonance (SPR) analysis, binding seen on the array was confirmed, and a binding constant as low as Kd ) 2.02 × 10-6 M was measured. Introduction The study of small molecule-protein interactions has become biologically relevant not only for the functional understanding of protein modulation but also in the discovery of novel protein ligands.1 Traditional chemical genetics approaches often involve time-consuming screening procedures in which library members are screened individually and complex pull-down experiments are required.2 To accelerate ligand discovery, a robust method incorporating small-molecule immobilization on microarrays in which thousands of small molecules are covalently immobilized on glass substrate platforms that provide for miniaturized, highthroughput screens against any potential binder was thereby conceived.3 Along with the need for sufficiently large, highly pure compound libraries, the vast potential of microarray technology intrinsically hinges on array fabrication that demands creative immobilization and surface modification strategies in order to ensure uniform slide printing, thereby increasing the veracity and reproducibility of downstream results. Herein, we describe the rapid generation of a small-molecule microarray for the efficacious discovery of ligands of a candidate protein, human IgG, (with Kd values as low as 10-6 M, as measured by surface plasmon resonance, SPR), thereby highlighting the value of a truly addressable small* To whom correspondence should be addressed. E-mail: yt.chang@ nyu.edu. † Department of Biological Sciences, National University of Singapore. § Department of Chemistry, National University of Singapore. ‡ New York University. ⊥ Both authors contributed equally to this work.

molecule microarray (SMM) and lead identification. The array produced herein is fully addressable in that our strategy synthesized the small-molecule library through a parallel route in which each compound’s unencoded identity was known before printing and easily identifiable, post-screening, merely from its position within the array. Schreiber et al. recently presented a microarray-based strategy using the socalled diversity oriented synthesis (DOS) in the discovery of calmodulin4 ligands and a transcriptional factor inhibitor.5 However, with the encoded, split-and-mix DOS library strategy adopted, every hit generated from the microarray screening necessitates a further deconvolution effort to determine its identity, thus limiting the aforementioned throughput, as promised by microarray-based technologies, especially where many putative hits arise. Results and Discussions Triazines (1,3,5-triazine) possessing 3-fold symmetry were chosen in our library as the molecular scaffold because they allow for versatile modifications uncomplicated by regiochemcial worries and have proven themselves to be useful biological targets.6 To further ameliorate the task of developing purposeful leads, members of the library were engineered with one of a range of amino-functionalized linkers that serve as a uniform, rapid, covalent attachment point to glass slides derivatized with N-hydroxysuccinimide esters.7 We had previously reported the synthesis of libraries of triazine compounds through our orthogonal solid-phase synthesis pathway using various synthetic strategies.8 Herein, we report a robust, tagged triethylene glycol (TG)-based amino linker triazine library (TG library) that incorporates

10.1021/cc049900s CCC: $27.50 © 2004 American Chemical Society Published on Web 09/03/2004

Discovery of Small-Molecule Ligands of Human IgG

Journal of Combinatorial Chemistry, 2004, Vol. 6, No. 6 863

Scheme 1. General Synthetic Scheme for Construction of TG Triazine Library

a new orthogonal synthetic pathway, affording much greater functional group diversity over our previous tagged library efforts9 (Scheme 1). The tagged linker strategy is desirable because incorporation of the linker from the beginning, before biological screening and as an intrinsic part of the compound, eliminates worries about any downstream tether effect, thus providing a straightforward means for target isolation without any detrimental effect to the compound’s activity or the need for further structure activity relationship (SAR) experiments.9 This TG library compliments and offers advantages over our previously reported efforts8,9 in a couple of ways. The synthetic scheme differs in that a mono (amino or hydroxy)-

substituted cyanuric dichloride moiety was loaded onto a TGlinker-functionalized resin, whereas previously, a linker mono-substituted (the linker as the first substituent) cyanuric dichloride was loaded onto the resin as the first step in the solid-phase synthesis (Scheme 1).9 This TG-linker-functionalized resin allows for rapid library diversification through simple splitting of the resin. As a consequence of the altered scheme, the second and more important improvement is the addition of primary alcohols to the library building block palette that were unattainable with our previous approach. Primary alcohols may only be efficiently and cleanly added to the cyanuric chloride scaffold as the first (of three) substitutions. This is due to the drastic decrease in reactivity

864 Journal of Combinatorial Chemistry, 2004, Vol. 6, No. 6

Uttamchandani et al.

Scheme 2. General Experimental Schemea

a (a) Directed immobilization of triazine libraries to generate a high-density microarray. (B) Incubation with a fluorescently labeled protein. (C) Removal of the unbound protein through washing. (D) Detection with instantaneous deconvolution of positive hits. (E) Assessing efficacy of hits using SPR.

seen with substituted cyanuric chloride analogues. Introducing an alcohol moiety as the first substituient, thus forming a building block II which can be subsequently loaded to a TG-linker-functionalized resin, is a very useful addition to our chemical toolbox and allows for N versus O atom substitution comparisons with hit compounds in later studies. The general tagged linker strategy is advantageous for a number of additional reasons. The basic linker used in all cases, 2,2′-[1,2-ethanediyl-bis(oxy)]bisethanamine, is commercially available and affordable and is easily monoprotected (N-Boc) in one step. Compound cleavage from the resin and linker deprotection is accomplished simultaneously in one step. The linker provides a sufficient space between the compounds and the microarray surface, at the same time allowing for greater conformational flexibility in the immobilized compounds. Furthermore, its hydrophilic character may provide a more protein-friendly environment during subsequent microarray screening processes. Last, the amino functional group allows for facile small-molecule immobilization and for a rapid transition to further downstream studies, such as affinity matrix pull-down experiments, without the need for any hit compound modification. The compounds were spotted, in duplicates, as an SMM on a modified glass substrate derived from standard microscope slides in a deterministic fashion that ensures immediate high-fidelity locus-based identification (Scheme 2). In total, 5376 spots corresponding to 2688 triazine-based library compounds were printed; 1152 of those were TG compounds and were synthesized as reported herein, and 1536 compounds were synthesized as reported previously by our group.9 In addition, we included in our arrayed grids a dye reference to not only validate the slide derivatization process, but also appropriately home in the software in the subsequent data acquisition. Immunoglobulins have seen numerous applications spanning immunoassays, diagnostics, and immunotherapeutics.1c The production of immunoglobulins, for example, valuable

humanized variants, for therapeutic applications requires stringent purification measures before being administered as approved drugs. However high molecular weight ligands, such as staphylococcal protein A and streptococcal protein G, are unfavorable for medicinal applications for their potential pyogenic effect as well as for other problems, including low biological stability, leakage from solid support, and difficulty in large-scale production and purification, contributing to high overall cost.10 Recent literature has shown that triazines may prove useful small-molecule ligand alternatives to IgG. For example, Li et al. used computeraided molecular modeling to successfully identify triazene analogues that bind to IgG with affinity constants of 105106 M-1.1c We thus hypothesized valuable potentials in screening human IgG against our arrays not only as proof of our overall concept but also in the discovery of efficacious ligands with direct relevance to industry. Human IgG was fluorescently labeled with Cy3-NHS to allow for sensitive visualization of small molecule-IgG interactions on the array. Spotted slides alone, without incubation with labeled IgG, were also scanned to ensure that the fluorescence did not originate from the spotted compounds themselves. The resulting scans were typical to that seen in Figure 1. Cases in which only one of the two duplicate spots displayed a substantive signal were dismissed as artifacts, and only hits that corroborated well in repeated experiments were deemed true positives. Three of the strongest hits on the array, based on intensity, were chosen for further validation, namely AMD10, AMD3, and K28. A faint positive, K42, and a negative, APF29, were also used as comparative benchmarks. In separate control experiments, other fluorescently labeled proteins (unrelated to human IgG) were used to screen against the same slide: none of the hits (e.g. AMD10, AMD3, K28, and K42) showed any positive binding, indicating that their binding toward human IgG is, indeed, highly specific. The spot intensities

Discovery of Small-Molecule Ligands of Human IgG

Journal of Combinatorial Chemistry, 2004, Vol. 6, No. 6 865

Figure 1. NHS-derivatized slides with 2688 triazine compounds spotted in duplicates and probed with human IgG-Cy3. The actual-sized array is enclosed in a blue box, with blow-ups describing the loci and the corresponding molecules that were selected for further assessments. (a-c) Correlated with strongly positive molecules, producing spot intensities at least two times that of the background. An intermediate (d) and a negative control (e) were also picked for comparative assessment. The reference control (f) is shown, and four sets of the Cy3-NH2 dye were printed at the ends of the grids.

Table 1. Microarray and SPR Results Obtained with Five Selected Triazines small molecule AMD10 AMD03 K28 K42 APF29

array signal (fluorescence units)

Kd/M

χ2

179 (++) 185 (++) 143 (++) 65 (+)